Local decrease of entropy, does it require life? Universal entropy can decrease only locally at the expense of bigger increase elsewhere.
Can this occur in a lifeless environment or does it necessarily require living organisms to do it?
Can this occur spontaneously or does it have to be an intentionally arranged process, like building a refrigerator?
My assumption is that you need to spend purposeful effort to decrease entropy locally. You need to spend energy to create differences in energy density and you need to have a reason why you do it. Living organisms use energy to create and maintain their internal order for the reason of survival. Inanimate matter has no reason to do anything. Causal uncontrolled processes always go towards higher entropy.
This question seems to enter the grey zone between physics and philosophy. Does a local decrease of entropy require intentional control over the course of events?
 A: It happens about 50 times in a second in any internal combustion engine, including cars. Without life. Life has no specific meaning in thermodynamical sense.
A: When something hot cools off it loses entropy locally, and its environment gains it.
When an animal eats food and emits heat and waste products, it loses entropy locally, and its environment gains it.  The food is converted into higher entropy waste products, and the extracted low entropy is used to maintain the animal.
When an internal combustion engine burns fuel, it loses entropy locally, and its environment gains it.  The low-entropy fuel is converted into useful low entropy work, and the higher entropy byproducts are emitted.
All forms of "entropy lowering" require emitting high entropy "waste".  To locally reduce entropy, you need to add in low entropy and convert it to waste.
Lava flowing into an ocean.  It has lots of heat (entropy), which it loses to the water producing steam (cold water is low entropy input, steam is high entropy waste).
Initial  <-- low entropy input
System   --> high entropy waste
 \/        
Lower entropy system

I suspect the above is what you are getting at.  Life tends to do the above a lot.
The trick is that the transfer of entropy from the low entropy input to the higher entropy waste is greater than the lowering of entropy of the system itself.
The game then becomes assigning what is the input, what is the waste and what the system is.
Life, and our inventions, tends to have a clear macro-state "system" you can talk about.  But the same thing can happen on molecular scales.
If you have any chemical reaction with two components A and B and two outputs C and D, such that C is lower in entropy then A, it would qualify.  To be recognizable, the reaction would also have to cause C and D to separate (like one becomes a gas, the other does not).
A: No life needed for this. All you need is for heat to flow away from the local region. It will carry entropy with it. Example: make yourself a cup of coffee. Put the cup on a table and wait while it cools. The entropy of the cup of coffee falls (and the entropy of the surrounding air increases).
